Building with AI / LLMs

Three artifacts make this SDK legible to coding assistants and agents:

llms.txt

/llms.txt is a one-line-per-page index of this entire site — path plus a one-sentence summary — regenerated with the documentation. Point a coding assistant at it to let the model pick the right page instead of crawling.

MCP server

The MCP server exposes every market-data endpoint to any Model Context Protocol client over JSON-RPC, so an LLM can pull real data mid-conversation. Setup is one config block; see the MCP page.

OpenAPI specification

/thetadatadx.yaml describes the local server's HTTP surface in OpenAPI 3: every route and parameter, and the response envelope. The columns of each route's rows are listed on its reference page.

WARNING

LLM output varies run to run. Treat generated queries and generated analysis as drafts: check parameters against the reference pages and validate results before acting on them.